GPS Transitions to New Control System 170
gsfprez writes "It took us a long time, but the Air Force has finally moved off of the 1970's mainframe GPS control system and is now running on a new Unix-based Control System called AEP — Architecture Evolution Plan. It's important to remember that current GPS satellites are basically solar powered iPod shuffles with atomic clocks that simply playback whatever we upload into them at a precise rate. They don't actually have any idea where they are — its the control system at Schriever Air Force Base that does. The new system will be a lot cheaper to support and modify since Sun stocks things like SATA drives - while digging up Saturday Night Fever-era DASDs isn't simple. AEP will also allow us to be ahead of the curve: we're basically good to go to fly the new IIF birds."
Maybe My Imagination (Score:2, Interesting)
Re:Wait... only one base providing data refersh? (Score:3, Interesting)
Chip H.
Re:i was edited and my points were lost.... (Score:4, Interesting)
Hope the reliability is just as good! (Score:5, Interesting)
However, I wonder who's handling the conversion for them, or if the Air Force is doing it themselves. I've seen great legacy conversion projects, and been involved in some really awful ones. One problem is just a lack of people who know enough about the "old" system to implement the software in the "new" side. The other, and far worse one is when companies (not militaries, mind you) bring in contractors who know _nothing_ about the hidden surprises in the old system, or nothing about the actual real-world application the computer is supporting.
As long as the system's not running J2EE or outsourced to a bunch of "expert" consultants, I'm guessing we're fine. But there is one key thing that's lost on "modern" IT -- proven systems work. Just because something is new doesn't mean it will work better! This is why I'm glad they stuck with UNIX instead of Linux or Windows.
Side note, how much do you think IBM was charging to maintain that monster??
Re:Hope the reliability is just as good! (Score:4, Interesting)
The 2nd Space Operations Squadron and the 19th Space Operations Squadron at Schriever Air Force Base are the primary operators of GPS. Within the squadrons, you have a wide variety of expertise -- airmen, government civilians, and contractors from the companies that developed both the new ground segment and the satellites that are on-station. Some of them are two-stripers just out of technical school
Re:i was edited and my points were lost.... (Score:2, Interesting)
I've worked on mainframes in the last few years. IBM is very much in the business of updating and supporting them. There are new versions of z/OS and z/VM that are as up to date and much more feature rich and reliable than most other platform OS. I guess another point I would make is that there is no better clustering available for reliability and geographical disparity than SYSPLEX in IBM z/OS. We have CECs in US and UK, and elsewhere, and all run off the same clock, back each other up for availability, and have dynamic pathing to various disk and tape and virtual devices that are available to any user, on any part of the system at any time. The closest thing to sysplex may have been VAX clustering, and there still is no unix that approaches that (go ahead, say it ain't so, then let me see your resume of mainframe and VAX work).
I can't see moving to unix unless it was either a loss of knowledge (retirement, etc.) or from a simple political gain perspective (look at me!! I "harvested" the legacy system and moved us to newer stuff!! I should get promoted!). Mainframes are very much updated and vital to computing today. Everyone has this erroneous idea that mainframes are all old, dusty machines that just run obscure code that some guy wrinte in '54... Wrong, the z in zSeries stands for ZERO-downtime. WE just did a complete processor and memroy upgrade on the fly into our US CECs last week (turned on 2 processors and added 2 gig of RAM). No downtime, no productivity loss, complete transparency to users. Please, find me a unix box that will do that. The perception that mainframes are rusting while the rest of the world moves on is a fallacy, and the idea that the machines cost too much for their function is simply a function of the great BSD argument... IF you want a driver for a $5 tape drive, then you have deemed your data worth $5. Reliability, scalability, and availability are really the givens in teh mainframe world that other vendors push so hard to tell you they have. The mainframe was literally designed for 99.999% uptime, I think that's about 3-7 minutes of downtime per year (if memory serves at all). I remember one of our older guys telling us about their shop in the early 90s that promised to fire the manager of their mainframe staff for 10 minutes of unplanned downtime in a year, and the manager felt that to be a reasonable reaction, he expected 3-5 minutes.
The software is updated, the machines are cutting edge (look for zSeries on ibm.com), and the I/O systems are state of teh art. Mainframes ahve been doing dynamic pathing SAN for about 10-15 years (I believe). Some SANs are just coming into the director idea of path management, IBM has used SAN-like technology for quite some time, and allows for dual writes to seperate devices (PPRC) and to writing to non-volatile RAM in teh controller and returning the block on the channel with the write to RAM. I realize this is similar to current systems and SANs, but the redundancy and pathing are not things I find on streetcorners. Have a lookat the new mainframes, and some of tha articles posted here about linux uner z/VM, they are amazing machines...
Later,
FC
What problem does this fix? (Score:2, Interesting)
Upgrading from 1970s technology to Unix? Unix *is* from the 1970s! The whole reason most slashdotters think it's the whole world is because they grew up with it -- i.e. it's "always" been here. OK it's been updated a lot since the old days but so have IBM mainframes. DASDs are SCSI disks these days.
Sorry to rant, I'm just so sick of companies/governments pouring resources into replacing working systems just because the current crop of wet-behind-the-ears CS grads have been trained to snicker at the stuff that has been making everything work like clockwork all these years.
Re:What problem does this fix? (Score:2, Interesting)
So it works _now_ on the mainframe and that's great. What happens when it breaks? Who's going to fix it if no one has expertise on that dinosaur of a system anymore? Someone who charges a very, very high rate, no doubt, because their skills are exceedingly rare. Not to mention getting parts for it if hardware breaks.
Its the same scenario that we all deal with, with our home PC/Mac systems. Sure we could all surf the net with 10 year old Pentium PC's. But at some point the cost to:
-Find and patch old software
-Buy additional 4 GB hard drives when we run out of space
-Replace Optical drives
-Replace burnt out fans
-etc
Plus the cost of our wasted time waiting for old equipment. Eventually, maintaining the old exceeds the cost of purchasing new. This happens with almost every piece of equipment. Cars, tractors, computers, houses, clothes, shoes, etc.
We've made astounding progress in technology in the past 20+ years. Just because something old gets the job done doesn't mean we can't do the same thing a whole lot better/faster/cheaper today with current technology.
Re:Wait... only one base providing data refersh? (Score:4, Interesting)
Probably not with either - as the ground control system is pretty big and delicate [1], pretty power hungry, and requires a fair number of specially trained personell [2] to operate it. It isn't something you are going to do in the back of a Humvee or a Bradley. You'd be hard pressed to do it in much of anything mobile short of the a Tico or a CVN.
That being said, the current generation of GPS birds are designed to operate autonomously for (IIRC) at least a month, though it will be some years before the entire constellation is upgraded to that standard. This implies the existence of a 'cold' backup somewhere else.
Insofar grandparents concern about 'high energy nukes' goes... He's pretty much out to lunch. The GPS constellation isn't as vulnerable to EMP/radiation effects as 'normal' LEO birds are because a) they are designed to be resistant to EMP, and b) the GPS constellation isn't inside the inner Van Allen belt like the birds wrecked by Starfish. You are pretty much in the situation of having to, even with nukes, take out each bird individually. (Sometimes they are close enough that you might be able to get 3-4, but the constellation is redundant enough that this won't take the system down.) So you are talking a pretty expensive and hard to hide endeavor, and being unable to take down enough of the cluster in a short enough timeframe to hamper US operations... before your own country is a glass parking lot.
I know many Slashdotters may have a hard time believing this - but they did actually think this stuff through when they designed the system.
[1] It's not just computers, but communications systems, precise clocks, etc... etc...
[2] Not just the techs that maintain the hardware above but the analysts that work with the incoming data to generate the corrections.